Six relevant RCTs were identified (n=4,672 patients, range 104 to 1,692). Jadad scores were relatively low (four trials scored less than 3 points, and the remaining two trials scored 3 and 4), partly because three trials used an open-label design and one provided little relevant information; two RCTs were double-blind.
The network analysis found a significant effect for docetaxel (HR 0.85, 95% CI 0.72 to 1.00), erlotinib (HR 0.71, 95% CI 0.58 to 0.85), and gefitinib (HR 0.88, 95% CI 0.78 to 0.99) versus placebo but no significant effect for pemetrexed versus placebo (HR 0.85, 95% CI 0.65 to 1.08), with one RCT for each comparison. There was no significant difference in effect for gefitinib versus docetaxel (two RCTs) or pemetrexed versus docetaxel (one RCT) or for an estimated hazard ratio for erlotinib versus docetaxel (mean HR 0.83, 95% CI 0.65 to 1.06). A pairwise meta-analysis was possible for gefitinib versus docetaxel, but the result was not significant.
The limited network analysis found a stronger significant effect for docetaxel versus placebo (HR 0.51, 95% CI 0.24 to 0.96) and a similar significant effect for erlotinib versus placebo (HR 0.71, 95% CI 0.58 to 0.85).
Network meta-analysis results were also displayed as probability of treatment occupying different rankings; there was an 85% probability that erlotinib was the most effective treatment.